BMJ  2004;329:1051 (30 October), doi:10.1136/bmj.329.7473.1051

reviews

PERSONAL VIEWS

Does evidence based medicine do more good than harm?

Advocates of evidence based medicine (EBM) have unleashed a huge move towards asking explicit questions, critical appraisal of the literature, finding and applying the best available evidence, and a rigorous quantitative approach to medicine. But does EBM do more good than harm?

If we argue that medicine needs to be evidence based, then logically we need evidence to support EBM. I have yet to find that evidence. It would be impossible to design a randomised study evaluating the cost effectiveness of EBM that would fulfil the criteria for qualifying as "evidence," because of the contamination that occurs from one case (doctor) to the next and because the sample size would need to be enormous. Probably the only feasible study would be to measure the costs and (subjectively experienced) usefulness of different forms of continuing education that use an EBM approach and to compare them with a regular approach.

Do we even need evidence that evidence based medicine works?

Do we even need evidence that EBM works? Some people say that the evidence in favour of using EBM is the evidence itself. Isn't it obvious that we should be well informed before we make a medical decision? I sometimes wonder why we needed a new movement to make this point. However, what EBM teaches is a set of skills to keep us up to date with the literature—and it is up to each one of us to adapt these skills to our personal needs. The steps of EBM should be considered aids and the checklists guides to help us think critically about what we read, but at the same time we need to remain critical of the approach itself.

The hierarchy of evidence suggested by EBM may not be justified and can be misleading. For example, a systematic review of a few small, poorly conducted, open randomised controlled trials is clearly not better than one large, well done, double blind trial. The hierarchy may not apply to the problem at hand. A randomised controlled trial is not the best way to determine rare side effects of a treatment: a case control or observational study is better. Randomisation is not always ethically justifiable. We can't randomise patients with end stage lung disease to transplantation and no transplantation arms, but we can model a control arm on the basis of data collected from patients on the waiting list for a donor organ. Sometimes requiring randomisation as evidence before an intervention is ridiculous: no study would compare plaster casts with expectant management for a fracture, even though the use of plaster casts is based solely on observational data. Finally, even a randomised study done according to state of the art methods can lead to spurious ridiculous results: a randomised controlled trial showed remote retroactive prayer to be effective ( BMJ 2001;323: 1450-1[Abstract/Free Full Text]).

Research into causes of illnesses and prognoses is usually best done with cohort studies—lower in the EBM hierarchy of levels of evidence but vital to our understanding of disease. Evaluating diagnostic tests is most efficiently done with a cohort study and decision model; a randomised trial of diagnostic tests is justifiable only under certain circumstances. In fact, every type of study has a place, even the case report. The case report is where it all begins: the equivalent of the time honoured method of teaching doctors during grand rounds. Published case reports focus on the rare and unexpected, but it is precisely these unexpected observations that lead us to question our beliefs and that can lead to new ideas and developments.

Besides the negative effect that EBM can have on how we appraise the literature, we may waste resources through inappropriate research, especially randomised controlled trials, by blindly conforming to EBM's levels of evidence. Meanwhile patients are denied potentially beneficial treatment. We should make the best decision according to the best available evidence and not withhold treatment just because we lack a randomised controlled trial and a systematic review. Value of information analysis can show whether doing more studies on a particular problem will reduce uncertainty to the extent that the research costs are justified. If they are not justified the resources would be better spent elsewhere.

We need to keep in mind that EBM reflects a particular perception of how medical decisions ought to be made. Rational, quantitative decision making may be important—but patients and doctors are human beings, and human beings are by no means always rational. The most rewarding aspects of caring for patients are neither rational nor quantitative. Focusing too much on the rational and quantitative aspects of clinical problems—an inherent danger in EBM—can have a negative influence on the doctor-patient relationship and can erode the caregiver's role in providing "care" in the fullest and most human way possible. Patients need empathy and understanding in order to express their preferences, values, and fears. Evidence is not enough: we need to communicate with our patients, listen to their concerns, elicit their values, be involved, really care about them. We also need to integrate the evidence with patients' values and preferences. Caring for patients is a complex process: it requires evidence, critical thinking, communication, judgment, intuition, and—most of all—tender loving care.

---

Erasmus MC, Rotterdam, The Netherlands m.hunink{at}erasmusmc.nl

CiteULike    Complore    Connotea    Del.icio.us    Digg    Reddit    Technorati    What's this?

Related Articles

Evidence from samples of one

Astier M Almedom
BMJ 2004 329: 1052. [Extract] [Full Text]

Learning from evidence based mistakes

Hilda Bastian
BMJ 2004 329: 1053. [Extract] [Full Text]

Rapid Responses:

Read all Rapid Responses

Evidence-Based Investigation into the Relation Between Sexual Intercourse and Pregnancy

Jacob M. Puliyel, et al.
bmj.com, 10 Nov 2004 [Full text]